This is a matched records study that will attempt to verify the accuracy of the reporting of age on the death certificate among persons in the extreme aged population. A sample of about four thousand decedents around 1970 will be selected from death certificates filed in the states of Pennsylvania and New Jersey. These records will be linked with the listing in the 1900 census, conducted when the decedent presumably was a youth or young adult. The motive to exaggerate age attributed to the elderly would not have been present at this early period. Effort will be made to correct age on the death certificate on the basis of the 1900 census records. The "adjusted" distribution of deaths by age around 1970 so derived, will permit the computation of corrected death rates and life tables. These data may then furnish a basis for improving mortality statistics at older ages, and for evaluating the quality of census data. In addition, they may provide a better appreciation of the basic causes of the anomaly of the cross-over of nonwhite mortality to lower levels than white mortality at the upper age range.